09 · Tracking · 8 min read

Track proposals and learn from the result

Keep a small weekly record of fit, proof, Connects, replies, calls, wins, and the lesson to test next.

Start with the decision you want the record to support

A tracker is useful when it changes a decision. You may want to know whether you are choosing suitable jobs, using relevant proof, spending within budget, or losing momentum after conversations. Choose that question before adding columns. A sheet with many numbers can still leave you unable to explain what you should do differently next week.

Use one row for each application or opportunity and keep a stable link or identifier. Record the submission date, type of opportunity, relevant service, and a short reason it looked like a fit. Keep invitations and repeat-client opportunities distinguishable from cold applications because the starting relationships differ. Combining them can produce an attractive average that describes none of the actual workflows.

Write the fit reason before seeing the outcome. This reduces the temptation to decide that every successful proposal was obviously a good choice. Record mandatory requirements, the proof you intended to use, and any uncertainty. You can then review the quality of your selection without relying entirely on memory.

Define events before you count them

Create a short definition for each stage in your own tracker. Submitted means the proposal was actually sent. Viewed means the platform recorded a view where that information is available. A substantive written reply, a scheduled call, a completed call, an estimate sent, and an active contract should be separate events. Use dates so an opportunity can move through several stages without losing its history.

Upwork's Stats and trends guidance currently describes Interviews as proposals that led to client responses and Hires as proposals that led to a contract. It also distinguishes boosted and organic proposal activity. Preserve those platform labels when copying a dashboard. Do not relabel Interviews as attended video calls because that would change the meaning of the data.

Use “unknown” when a field is unavailable and “pending” when an outcome is unresolved. Neither is the same as zero or rejected. If a client withdraws a project, record the observable fact rather than guessing that your price or writing caused it. Consistent definitions matter more than a complicated dashboard.

Compare opportunities from a consistent period

Choose whether you are analysing proposals submitted in a period or events observed during that period. These answer different questions. A reply this week may belong to a proposal from last month. If you divide this week's replies by this week's submissions without matching the records, the result may not describe the performance of either group.

For a submission group, record the date range and the date you last checked outcomes. Give later replies time to appear. Keep a note of proposals that are still pending rather than quietly treating the latest group as fully finished. When comparing periods, use similar observation windows so a mature set is not compared with applications sent yesterday.

Keep important differences visible: service category, job fit, price range, invitation versus application, and boosted versus organic activity. You do not need a separate report for every attribute. Start by identifying the largest difference that could explain the result. With a small sample, a descriptive note is often more honest than a precise-looking percentage.

Track spending and time alongside activity

Record Connects used for the application separately from any additional boost spend. Use the actual account record when available, especially if Connects were returned or a promotion affected the balance. Upwork currently lists purchased Connects at US$0.15 each, but a standard purchase price is not automatically your cash expense for every application. Distinguish consumption from purchases and subscription allowances.

Record enough time to understand the cost of finding work: researching jobs, preparing proposals, handling replies, and attending calls. You do not need to time every sentence. A consistent, practical estimate is better than ignoring the time entirely. State whether your cost calculation includes only external spending or also the value of your own time.

When a contract begins, separate booked value, money earned, and net proceeds. A signed project amount is not the same as collected revenue or profit. Platform fees, delivery effort, refunds, and other costs affect the eventual result. If your tracker cannot support a profitability claim, report the stage it does support instead.

  • Application Connects and any boost spend.
  • Actual purchases or membership costs, kept separate from token usage.
  • Research, writing, conversation, and call time.
  • Contracts started, amounts earned, and costs recorded as different measures.

Read the funnel as a set of hypotheses

If suitable proposals are not viewed, inspect job selection, mandatory requirements, submission completeness, and the opening. The client may also have stopped reviewing applications. A missing view does not prove that one sentence caused the problem. Note what you can inspect directly and what remains outside your evidence.

If viewed proposals do not lead to useful replies, review whether your proof matches the problem, your role is clear, and the next step is easy to understand. Price, competition, timing, and the buyer's internal plans can also matter. Avoid automatically lowering rates after a quiet period. First identify which explanation you can test without making a large unnecessary change.

If conversations stall, review the unresolved decision. Perhaps the technical scope is unclear, the buyer has no budget approval, or the next action has no owner. If calls happen but contracts do not, inspect the recap and quote alongside the call notes. Treat these as diagnostic questions, not universal explanations supplied by a funnel chart.

Run a small test you can interpret

Choose one change that addresses a specific observation. You might narrow job selection to work your existing sample demonstrates, answer a frequently missed screening question more clearly, or improve the first two lines of a proposal. Write down why you expect the change to help and what evidence would make you reconsider.

Keep other major choices reasonably steady while you observe the next comparable group. If you change niche, rate, profile, proof, and bidding behaviour together, you may see a different outcome but learn little about the cause. Real marketplaces cannot be controlled like a laboratory, so avoid claiming scientific certainty from a small personal experiment.

Decide the review point in advance and respect your spending limit. Record the number of relevant opportunities, the observation window, and any unusual conditions. A few replies can suggest a useful direction, but they do not establish a dependable conversion rate. If the evidence is weak, keep the conclusion modest and choose the next question carefully.

Fill-in experiment record — not a client result

Observation: [specific pattern]. Change: [one action]. Comparable jobs: [criteria]. Budget and review date: [limits]. Outcome observed: [counts with definitions]. Unknowns: [unresolved factors]. Next decision: [continue, revise, or stop].

Read the public UGD snapshot as one historical record

The saved UGD account view labelled 2025 shows 244 proposals sent, 85 viewed, 44 interviews, and 13 hires. You can inspect the original image on the public evidence page. The accompanying text distinguishes those recorded counts from UGD's reported work around job selection, proposal drafts, conversations, and reporting.

Those counts do not establish revenue, profit, improvement over a baseline, or which action caused a hire. They are not a prediction for another freelancer. The image also does not provide all the row-level information needed to analyse every proposal's timing and relationship to the other events. Preserve that limitation when using it as a teaching example.

The useful lesson is how to label a record: name the source, preserve the period and metric names, explain the provider's role separately, and state what the evidence cannot answer. Apply the same discipline to your own reports. A transparent account of uncertainty is more useful than turning a historical screenshot into an unsupported promise.

Use a short weekly review and protect the underlying data

Set aside a regular review time. First check that the records are current. Then identify where progress stopped, read a few actual proposals or conversations, and choose one action for the next period. Finish by confirming the spending limit and any follow-up commitments. This keeps the tracker connected to work rather than turning it into a separate administrative project.

Keep client secrets and unnecessary personal data out of the sheet. A link and a brief factual note may be enough. Restrict access to people who need it, and do not paste private conversations into an unapproved analysis tool. When a public example is useful, create a separate version with permission and remove details that are not needed to explain the point.

Keep a simple change log for your own process. Note when you changed the offer, profile, sample, or pricing approach. Those changes may affect later comparisons. Revisit your column definitions when confusion appears, but preserve older records accurately. The objective is a record you can trust when deciding where to spend your next hour and your next Connects.

Questions people ask

Clear answers before you act.

What is a good proposal response rate?

There is no single rate that fairly describes every service, account, budget, and period. Define the events and compare suitable opportunities with similar observation windows before drawing a conclusion.

Should unanswered proposals count as rejected?

No. Keep unresolved proposals pending unless you have evidence of a decision. A missing response does not tell you why the client did not proceed.

Do I need a paid CRM?

A simple sheet can be enough when it has stable definitions, clear dates, and an accessible review routine. Choose a more complex system only when it solves a real workflow problem.

Continue with a focused guide

Sources and corrections

Check the source, not just our word.

Platform details can change. This guide was reviewed on 18 Nov 2026. If something looks wrong, email team@upgrowthdesk.com.

About the author

Abhishek Bhambhana

Abhishek is the founder of UpGrowthDesk. He works on Upwork growth, business development, client conversations, and simple systems for freelancers and IT teams.

Meet the founder Ask about help with this
Put it to work

Use the guide on a real task.

Start with the free tool, or see the complete service when you want a person to help.